Applicability.

1.  Nondividing cells (neurons, macrophages, other cells that cannot be transduced with traditional retroviral vectors).  Infections in vivo (injection into an animal) and in vitro (dissociated cultures, organotypic slices) work very well.
2.  Long-term infection.  Retroviral infections are not cytopathic, so the infection can be followed over a course of weeks or more.
3.  Low levels of expression.  Massive overproduction of protein, as is seen with the high-copy number cytopathic viruses, will not occur.
4.  Rapid testing of multiple constructs or genes of interest.  Creating a new construct is as easy as subcloning into a plasmid.

Lentiviruses are not recommended for:

1.  Dividing cells.  Traditional retroviral vectors, such as MMLV, are easier to work with, easier to obtain, and many packaging cells lines are available.  Don't use a lentivirus unless you need one.
2.  Those desiring to produce a large amount of protein.
3.  Acute slices, unless you have no other choice of vector.  Lentiviruses are hard to produce in the titers required for picoinjection, and express too slowly to be ideal for the 24-hour maximum life of the slice.

Background and Choice of Systems.

The lentiviruses are a family of retroviruses capable of transducing non-dividing cells.  Replication-deficient vector systems based on lentiviruses have been developed by several laboratories:   those derived from HIV-1 are the most common (1,2), but there is also a vector system based on HIV-2 (3) and one on the Feline Immunodeficiency Virus (FIV) (4). HIV-based vectors can be missing the virulence factors vif, vpu, vpr, and nef and still transduce quiescent cells efficiently (5); such vectors can be used under BSL-2 conditions and are appropriate for researchers desiring to perform physiology experiments on unfixed, living infected cells.   This multiply-deleted system is available from F. Gage at the Salk Institute or from D. Trono at the University of Geneva; it may soon be available commercially.  This vector, as well as its precursor that contains all of the virulence factors and must be used under BSL-3 conditions, are the only lentiviruses being used at Caltech.  The vector comes as three DNA plasmids: one provides all of the necessary enzymes and structural proteins in trans; one contains the packaging signal for encapsidating the resulting RNA into virions; and the other provides the env protein for the desired pseudotype.  Env plasmids encoding both VSV-G and the MMLV env are available.  Click here for an illustration of each plasmid and a description of its function.
 
Cloning Hints
 As viruses go, retroviruses have small genomes.  But you still  have to work with plasmids of 9-15kb, which means there will be very few unique restriction sites.  There are several ways to engineer an insert that is easily subcloned into a retroviral vector:

1.  PCR.  The drawbacks are that the resulting product must be sequenced, and is prone to errors especially for large inserts.
2.  Artificial linkers.  Any restriction site can be made to fit any other by means of an oligo whose sticky ends match those of the two restriction sites.  For example, if your insert has BamHI ends and the viral vector cuts with PacI, design two oligos with the ends

After annealing the oligo, perform the3-way ligation of vector+insert+linker with a huge molar excess of linker.  Since it is not phosphorylated, it will not stick to itself and there will not be any concatomers.

3.  Blunt-ending.  Only do this if you have to.  This is much more work than the other two methods.

4.  Inserting a polylinker.  If there are restriction enzymes that don't cut the viral vector but which aren't included in its polylinker, you can cut in one of the sites and insert an oligo containing the restriction sites of your choice.  This becomes more and more useful as more rare-hitting enzymes are made available: good bets include PacI, AscI, and SwaI, all from NEB.  SwaI is blunt-ended.

Once the transfer vector of your choice has been created, it is necessary to transfect the three plasmids into packaging cells to generate the virus.  Large amounts of DNA will be required to produce useful amounts of virus: a Qiagen Maxiprep preparation is good for several calcium phosphate transfections; a Gigaprep may be more convenient.  Cesium-chloride purification or other ultra-pure DNA preparation methods are not necessary.  The plasmid pMD.G encoding the VSV-G protein grows to very low copy number in bacteria; at least 500mL of medium should be used to produce 1mg of plasmid DNA.
The packaging cells are 293T; these are derived from HEK 293 cells, with the further stable transfection of the SV40 large T antigen.  This allows a transfected plasmid bearing an SV40 origin of replication to be episomally replicated in enormous quantities: up to 100,000 plasmids per cell.  If the three plasmids are transfected into ordinary 293 cells, virus will be produced, but at very low titer: 103-104 particles/mL.  293T cells are available from many laboratories but are no longer sold commercially; as well as the Salk Institute, the cells may be obtained from Stanford (the Nolan lab).  They are maintained in DME high glucose supplemented with 10% fetal bovine serum, 2mM glutamine, and 100U penicillin/streptomycin; they may be frozen in 10% DMSO; and they are good for approximately 100 passages.  Cells too old for efficient virus production will grow at an unpredicatble rate, show morphological changes, and adhere badly to the tissue culture dish.
There is also a cell line available (293G) that is stably transfected with the VSV-G protein under control of the “tet-off” tetracycline inducible system.  The cells are maintained in the presence of tetracycline until virus is to be made, at which point tetracycline is removed from the medium and the G protein is expressed.  These cells grow much more slowly than 293T cells but remove the obligation to grow the pMD.G plasmid.  These are available from (???)
 

Transfection
A sufficient amount of DNA of the three plasmids, and 293T cells, are all that is needed to make virus.  293T cells can be seeded onto dishes for transfection the day before transfection or earlier: as they grow quite fast, it is necessary to watch them to make sure they don’t become overconfluent.  10-cm tissue culture dishes can be used, but 15-cm Integrid dishes (Falcon) are more convenient.  Three to six of these large plates is a typical production; one dish if the construct is new and not known to work.  Make sure that the dishes are labeled for tissue culture: most 15-cm dishes are intended for bacteria and are not coated for mammalian cells.  293T cells will not adhere to these dishes.
Calcium phosphate transfection is the standard method for transient expression of the three plasmids.  [Carlos]
The liposome-mediated transfection methods, including but by no means limited to LipofectAMINE (Gibco), Superfect and Effectene (Qiagen), are more convenient, require much less DNA, and are less toxic to cells than calcium phosphate.  The drawback is their expense: a 1- to 1.2-mL tube costs approximately $150; this will transfect 10-12 15-cm dishes.  The protocols for these reagents can be followed exactly; the amounts for 15-cm dishes are simply twice those for 10-cm dishes.  All of these compounds lead to high transfection efficiency and excellent titers.
 

References
1.  Naldini L, Blomer U, Gage FH, Trono D, Verma IM, Proc Natl Acad Sci U S A 1996 Oct 15;93(21):11382-8
2.  Corbeau P, Kraus G, Wong-Staal F, Proc Natl Acad Sci U S A 1996 Nov 26;93(24):14070-5
3.  Arya SK, Zamani M, Kundra P, Hum Gene Ther 1998 Jun 10;9(9):1371-80
4.  Poeschla EM, Wong-Staal F, Looney DJ, Nat Med 1998 Mar;4(3):354-7
5.  Zufferey R, Nagy D, Mandel RJ, Naldini L, Trono D, Nat Biotechnol 1997 Sep;15(9):871-5